Non-transformed cells require growth factors for their survival and proliferation. In addition to polypeptide growth factors, an emerging class of lipids with growth factor-like properties has been discovered, collectively known as phospholipid growth factors (PLGFs). Although they have similar pharmacologic properties in inducing the proliferation of most quiescent cells (Jalink et al., 1994a; Tokumura, 1995; Moolenaar et al., 1997), PLGFs can be sub-divided structurally into two broad categories. The first category contains the glycerophospholipid mediators (GPMs), which possess a glycerol backbone. Exemplary GPMs include LPA, phosphatidic acid (PA), cyclic phosphatidic acid (cyclic-PA), alkenyl glycerol phosphate (alkenyl-GP), and lysophosphatidyl serine (LPS). The second category contains the sphingolipid mediators (SPMs), which possess a sphingoid base motif. Exemplary SPMs include sphingosine-1-phosphate (SPP), dihydrosphingosine-1-phosphate, sphingosylphosphorylcholine (SPC), and sphingosine (SPH).
LPA (Tigyi et al., 1991; Tigyi and Miledi, 1992), PA (Myher et al., 1989), alkenyl-GP (Liliom et al., 1998), cyclic-PA (Kobayashi et al., 1999), SPP (Yatomi et al., 1995), and SPC (Tigyi et al., 2000) have been detected in serum. These lipid mediators have been identified and characterized. There are still yet unknown PLGFs present in the serum and plasma that exhibit growth factor-like properties (Tigyi and Miledi, 1992). LPA, at a concentration of ≈20 μM, is the most abundant PLGF present in the serum (Tigyi and Miledi, 1992; Jalink et al., 1993).
In eukaryotic cells, LPA is a key intermediate in the early stages of phospholipid biosynthesis, which takes place predominantly in the membrane of endoplasmic reticulum (ER) (Bosch, 1974; Bishop and Bell, 1988). In the ER, LPA is derived from the action of Acyl-CoA on glycerol-3-phosphate, which is further acylated to yield PA. Because the rate of acylation of LPA to PA is very high, very little LPA accumulates at the site of biosynthesis (Bosch, 1974). Since LPA is restricted to the ER, its role as a metabolic intermediate is most probably unrelated to its role as a signaling molecule.
LPA is a constituent of serum and its levels are in the low micromolar (μM) range (Eicholtz et al., 1993). This level is expected because LPA is released by activated platelets during the coagulation process. Unlike serum, it is not detectable in fresh blood or plasma (Tigyi and Miledi, 1992; Eicholtz et al., 1993). LPA that is present in the serum is bound to albumin, and is responsible for a majority of the heat-stable, and non-dialysable biological activity of the whole serum (Moolenaar, 1994). The active serum component that is responsible for eliciting an inward chloride current in Xenopus oocyte was identified to be LPA (18:0) (Tigyi and Miledi, 1992). The bulk of the albumin-bound LPA (18:0) is produced during the coagulation process, rather than by the action of lysophospholipase D (PLD) on lyso-PC. The latter pathway is responsible for the presence of LPA in ‘aged’ plasma that has been de-coagulated by the action of heparin or citrate plus dextrose (Tokumura et al., 1986). Furthermore, LPA is not present in plasma that has been treated with EDTA. This fact implies that plasma lysophospholipase may be Ca2+-dependent (Tokumura et al., 1986).
The role of albumin is to protect LPA from the actions of phospholipases present in the serum (Tigyi and Miledi, 1992). Tigyi and Miledi suggested that albumin not only acts as a carrier of LPA in the blood stream, but also increases its physiological half-life. There are yet unidentified lipid mediators present in serum albumin that mimic the actions of LPA in eliciting chloride current in Xenopus oocytes.
LPA-responsive cell types extend from those of slime mold amoebae and Xenopus oocyte to mammalian somatic cells. Thus, it seems likely that the source of LPA and its release may not be restricted only to activated platelets. Recent experiments showed that, on stimulation by peptide growth factors, mammalian fibroblasts rapidly produce LPA, which is followed by its release into the extracellular medium (Fukami and Takenawa, 1992).
There is evidence that relatively high amounts of bioactive LPA of unknown cellular origin are present in the ascitic fluid of ovarian cancer patients (Xu et al., 1995a), and that the ascitic fluid from such patients is known to possess potent mitogenic activity for ovarian carcinoma cells (Mills et al., 1988; Mills et al., 1990). It remains to be established whether it is secreted by tumor cells into the extracellular fluid, secreted by leukocytes, or produced from more complex lipids via the actions of various phospholipases.
GPMs and SPMs elicit a wide variety of cellular responses that span the phylogenetic tree (Jalink et al., 1993a). LPA induces transient Ca2+ signals that originate from intracellular stores in a variety of cells such as neuronal (Jalink et al., 1993; Durieux et al., 1992), platelets, normal as well as transformed fibroblasts (Jalink et al., 1990), epithelial cells (van Corven et al., 1989; Moolenaar, 1991), and Xenopus oocytes (Tigyi and Miledi, 1992; Durieux et al., 1992; Fernhout et al., 1992). LPA induces platelet aggregation (Schumacher et al., 1979; Tokumura et al., 1981; Gerrard et al., 1979; Simon et al., 1982) and smooth muscle contraction (Tokumura et al., 1980; Tokumura et al., 1994), and upon intravenous administration it induces species-dependent changes in blood pressure ((Schumacher et al., 1979; Tokumura et al., 1978).
LPA, when added to quiescent fibroblasts, stimulates DNA synthesis and cell division (van Corven et al., 1989; van Corven et al., 1992). The growth-like effects of LPA do not require the presence of peptide growth factors. This observation makes LPA different from endothelin or vasopressin, which require the presence of insulin or epidermal growth factor (Moolenaar, 1991) to sustain cell proliferation. Notably, in Sp2 myeloma cells, LPA demonstrated an antimitogenic response, which was mediated by an increase in cAMP levels (Tigyi et al., 1994; Fischer et al., 1998). Unlike the mitogenic pathway, the anti-mitogenic pathway was not affected by pertussis toxin (PTX). Also, on addition of forskolin and isobutyl methyl xanthin, the antimitogenic actions of LPA in Sp myeloma cells were additive (Tigyi et al., 1994). In various cell types, LPA causes cytoskeletal changes, which include formation of focal adhesions and stress fibers in fibroblasts (Ridley and Hall, 1992). LPA also promotes the reversal and suppression of neuroblastoma differentiation by inducing the retraction of developing neurites (Jalink et al., 1994a; Jalink et al., 1994b). Addition of nanomolar (nmol) amounts of LPA (Jalink and Moolenaar, 1992) to serum-starved N1E-115 neuroblastoma cells caused immediate neurite retraction, which was accompanied by rapid, but transient, rounding of the cell body (Jalink et al., 1993b). When a continuous presence of LPA is provided, neuroblastoma cells maintain their undifferentiated phenotype, but fail to undergo mitosis (Jalink et al., 1993b). Additional factors, such as insulin-like growth factors, are required for the progression of the cell cycle. Once the cells have undergone morphological differentiation, the addition of LPA reverses this morphological change. Thus, LPA-induced neurite retractions result from the contraction of the actin-cytoskeleton, rather than from loss of adhesion to the substratum (Jalink et al., 1993b; Jalink et al., 1994b).
LPA, similar to other physiological chemoattractants (e.g., interleukin-8), induces cell migration by a haptotactic mechanism in human monocytes (Zhou et al., 1995). In addition to inducing cell migration, LPA promotes the invasion of hepatoma and carcinoma cells into the monolayer of mesothelial cells (Imamura et al., 1993). The mechanism that underlies this invasion is still unclear, but it may be due to enhanced cell motility and increased cell adhesion. Finally, LPA is also known to block neonatal cardiomyocyte apoptosis (Umansky et al., 1997).
A unique natural phospholipid, namely cyclic-PA, was shown to be responsible for cellular actions that were similar to or opposite to other GPMs, depending on the cell type. When tested on the Xenopus oocyte, it elicited chloride current just like other GPMs; but its response was not desensitized by LPA (Fischer et al., 1998). Murakami-Murofushi et al. (1993) showed that cyclic-PA exhibited antiproliferative actions, unlike LPA, which induces proliferation.
PLGF receptors (PLGFRs) belong to a seven-transmembrane (7 TM) guanine nucleotide-binding regulatory protein (G protein)-coupled receptors (GPCR) superfamily. Seven-TM GPCRs are a family of cell-surface receptors that mediate their cellular responses via interacting with the heterotrimeric G-protein. A number of LPA receptors have been identified including, among others, EDG-2, EDG-4, EDG-7, and PSP-24. A phylogenetic tree illustrating the relatedness of these LPA receptors and others is shown in FIG. 1.
In 1996, Hecht et al. used differential hybridization to clone a cDNA encoding a putative serpentine receptor from mouse neocortical cell lines (Hecht et al., 1996). The gene was termed as ventricular zone gene-1 (Vzg-1). The gene was expressed in cortical neurogenic regions and encoded a protein with a molecular weight of 41 kDa (364 amino acids). Vzg-1 was very similar to an unpublished sheep sequence termed endothelial differentiation gene-2 (EDG-2). The same cDNA was also isolated as an orphan receptor from mouse and bovine libraries, and was known as rec1.3 (Macrae et al., 1996). It was widely distributed in the mouse tissue, with the highest expression in the brain and heart.
In 1996, Guo et al., using a PCR base protocol, isolated another putative LPA receptor PSP-24 (372 amino acids) from Xenopus oocyte (Guo et al., 1996). This receptor showed little similarity with Vzg-1/EDG-2/rec1.3 (Guo et al., 1996). A sequence based search for sphingolipid receptors, using the cDNA sequence of the EDG-2 human LPA receptor, led to two closely related GPCRs, namely, rat H218 (EDG-5, 354 amino acids) and EDG-3 (378 amino acids) (An et al., 1997a). Northern analysis showed a high expression of mRNA that encoded EDG-3 and EGD-5 in heart tissue.
The recent identification of EDG-2 as a functional receptor for LPA prompted An et al. to perform a sequence-based search for a novel subtype of LPA receptor (An et al., 1998a). A human cDNA, encoding a GPCR, was discovered and designated EDG-4 (An et al., 1998a). Northern blot analysis showed that, although EDG-2 and EDG-4 both serve as GPM receptors, their tissue distributions were very different. Unlike EDG-2, EDG-4 was primarily expressed in peripheral blood leukocytes and testes (An et al., 1998a).
PCR amplification cDNA from human Jurkat T cells identified a previously unknown GPCR belonging to the EDG family. The identified GPCR was designated EDG-7. It has a molecular mass of 40 kDa (353 amino acids). Northern blot analysis of EDG-7 expression in human tissues showed that it is expressed in heart, pancreas, prostate, and testes (Bandoh et al., 1999). Thus, there are two distinct families of PLGF receptors PSP24 and EDG, with a total of ten individual PLGFRs (FIG. 1). The list continues to grow.
These various receptors can be classified based on their ligand specificities for GPMs or SPMs, as shown in Table 1 below.
TABLE 1Phospholipid Growth Factor Receptor(PLGFR), Length and Principle LigandNumber ofPrinciplePLGFRamino acidsLigandEDG-1381SPPEDG-2364LPAEDG-3378SPPEDG-4382LPAEDG-5354SPPEDG-6385SPPEDG-7353LPAEDG-8400SPP XenopusPSP24372LPA MurinePSP24373LPA
Xenopus PSP24 and murine expressed PSP24 specifically transduce GPM-evoked oscillatory chloride-currents (LPA, Fischer et al., 1998). These are not structurally homologous to the EDG family (Tigyi and Miledi, 1992; Fernhout et al., 1992). The EDG family can be divided into two distinct subgroups. The first group includes EDG-2, EDG-4, and EDG-7, which serve as receptors for only GPM (Hecht et al., 1996; An et al., 1998a; Bandoh et al., 1999; An et al., 1998b) and transmit numerous signals in response to ligand binding. The second group involves EDG-1, EDG-3, EDG-5, EDG-6, and EDG-8, and is specific for SPMs (An et al., 1997a; Im et al., 2000; van Brocklyn et al., 1998; van Brocklyn et al., 2000; Spiegel and Milstein, 2000). Principle tissue expression of the various PLGFR's is shown in Table 2 below.
TABLE 2Human Tissue Expression of PhospholipidGrowth Factor ReceptorsPLGFRHuman Tissue with Highest ExpressionEDG-1UbiquitousEDG-2Cardiovascular, CNS, Gonadal tissue, GIEDG-3Cardiovascular, LeukocyteEDG-4Leukocyte, TestesEDG-5Cardiovascular, CNS, Gonadal tissue, PlacentaEDG-6Lymphoid, Hematopoietic tissueEDG-7Heart, Pancreas, Prostate, TestesEDG-8Brain PSP24 CNS
PLGFs activate multiple G-protein-mediated signal transduction events. These processes are mediated through the heterotrimeric G-protein families Gq/11, Gi/0, and G12/13 (Moolenaar, 1997; Spiegel and Milstein, 1995; Gohla, et al., 1998).
The Gq/11 pathway is responsible for phospholipase C (PLC) activation, which in turn induces inositol triphosphate (IP3) production with subsequent mobilization of Ca2+ in a wide variety of cells (Tokumura, 1995). In some cells, this response is PTX-sensitive, implying that there is involvement of multiple PTX-sensitive and insensitive pathways (Tigyi et al., 1996). This pathway is also responsible for the diacyl glycerol (DAG)-mediated activation of protein kinase C (PKC). PKC activates cellular phospholipase D (PLD), which is responsible for the hydrolysis of phosphatidyl choline into free choline and PA (van der Bend et al., 1992a). Also, PLC is capable of activating MAP kinase directly, or via DAG activation of PKC in some cell types (Ghosh et al., 1997).
The mitogenic-signaling pathway is mediated through the G-protein heterotrimeric Gi/0 subunit. Transfection studies indicate that the Giβγ dimer rather than the αi subunit is responsible for Ras-MAP kinase activation. The activation of Ras is preceded by the transactivation of the receptor tyrosine kinases (RTKs) such as EGF (Cunnick et al., 1998) or PDGF receptors (Herrlich et al., 1998). The transactivated RTKS activate Ras, which leads to the activation of MAP kinases (ERK 1, 2) via Raf. The Giα subunit, which is PTX-sensitive, inhibits adenylyl cyclase (AC), resulting in βγ dimer docking to a G-protein-coupled receptor kinase (GRKs) that phosphorylates and desensitizes the receptor. The phosphorylated receptor is recruited by β-arrestin, thus recruiting src kinase, which phosphorylates the EGF-receptor, generating its active conformation (Lin et al., 1997; Ahn et al., 1999; Luttrell et al., 1999). The transactivated RTKs, in turn, activate Ras, which leads to the activation of MAP kinases (ERK 1, 2) via Raf. The Giα subunit, which is PTX-sensitive, inhibits AC, resulting in decreased levels of cyclic-AMP (cAMP). The opposite cellular effects by LPA, that is, mitogenesis and antimitogenesis, are accompanied by opposing effects on the cAMP second messenger system. Mitogenesis is mediated through the Giα pathway, which results in decreased levels of cAMP (van Corven et al., 1989; van Corven et al., 1992), whereas anti-mitogenesis is accompanied by a non-PTX sensitive Ca2+-dependent elevation of cAMP (Tigyi et al., 1994; Fischer et al., 1998).
In contrast, very little is known about the PTX-insensitive G12/13 signaling pathway, which leads to the rearrangement of the actin-cytoskeleton. This pathway may also involve the transactivation of RTKs (Lin et al., 1997; Ahn et al., 1999; Luttrell et al., 1999; Gohla et al., 1998) and converge on a small GTPase, Rho (Moolenaar, 1997). Much more is known about the down-stream signaling of Rho because various protein partners have been isolated and identified. Rho activates Ser/Thr kinases, which phosphorylate, and as a result inhibit, myosin light chain phosphatase (MLC-phosphatase) (Kimura et al., 1996). This path results in the accumulation of the phosphorylated form of MLC, leading to cytoskeletal responses that lead to cellular effects like retraction of neurites (Tigyi and Miledi, 1992; Tigyi et al., 1996; Dyer et al., 1992; Postma et al., 1996; Sato et al., 1997), induction of stress fibers (Ridley and Hall, 1992; Gonda et al., 1999), stimulation of chemotaxis (Jalink et al., 1993a), cell migration (Zhou et al., 1995; Kimura et al., 1992), and tumor cell invasiveness (Imamura et al., 1993; Imamura et al., 1996). The PLGF-induced, Rho-mediated, tumor cell invasiveness is blocked by C. botulinium C3-toxin, which specifically ribosylates Rho in an ADP-dependent mechanism (Imamura et al., 1996).
Rho also has the ability to stimulate DNA synthesis in quiescent fibroblasts (Machesky and Hall, 1996; Ridley, 1996). The expression of Rho family GTPase activates serum-response factor (SRF), which mediates early gene transcription (Hill et al., 1995). Furthermore, PLGF (LPA) induces tumor cell invasion (Imamura et al., 1996); however, it is still unclear whether it involves cytoskeletal changes or gene transcription, or both.
By virtue of LPA/LPA receptor involvement in a number of cellular pathways and cell activities such as proliferation and/or migration, as well as their implication in wound healing and cancer, it would be desirable to identify novel compounds which are capable of acting, preferably selectively, as either antagonists or agonists at the LPA receptors identified above.
There are currently very few synthetic or endogenous LPA receptor inhibitors which are known. Of the antagonists reported to date, the most work was done on SPH, SPP, N-palmitoyl-1-serine (Bittman et al., 1996), and N-palmitoyl-1-tyrosine (Bittman et al., 1996). It is known that the above-mentioned compounds inhibit LPA-induced chloride currents in the Xenopus oocyte (Bittman et al., 1996; Zsiros et al., 1998). However, these compounds have not been studied in all cell systems. It is also known that SPP inhibits tumor cell invasiveness, but it is uncertain whether SPP does so by being an inhibitor of LPA or via the actions of its own receptors. N-palmitoyl-1-serine and N-palmitoyl-1-tyrosine also inhibited LPA-induced platelet aggregation (Sugiura et al., 1994), but it remains to be seen whether these compounds act at the LPA receptor. Lysophosphatidyl glycerol (LPG) was the first lipid to show some degree of inhibition of LPA actions (van der Bend et al., 1992b), but it was not detectable in several LPA-responsive cells types (Liliom et al., 1996). None of these inhibitors was shown to selectively act at specific LPA receptors.
A polysulfonated compound, Suramin, was shown to inhibit LPA-induced DNA synthesis in a reversible and dose-dependent manner. However, it was shown that Suramin does not have any specificity towards the LPA receptor and blocked the actions of LPA only at very high millimolar (mM) concentrations (van Corven et al., 1992).
The present invention is directed to addressing the lack of agents which act as LPA agonists and LPA antagonists.